Enable ruff's pandas-vet (PD) rules

[seisman]

Errors after enabling ruff's pandas-vet rules. Looks good to me but maybe we should ignore PD901?

examples/gallery/3d_plots/scatter3d.py:22:1: PD901 Avoid using the generic variable name `df` for DataFrames
   |
21 | # Load sample iris data
22 | df = pd.read_csv("https://github.com/mwaskom/seaborn-data/raw/master/iris.csv")
   | ^^ PD901
23 | 
24 | # Convert 'species' column to categorical dtype
   |

examples/gallery/histograms/blockm.py:28:1: PD901 Avoid using the generic variable name `df` for DataFrames
   |
26 | # Calculate mean depth in kilometers from all events within
27 | # 150x150 arc-minute bins using blockmean
28 | df = pygmt.blockmean(data=data, region=region, spacing=spacing)
   | ^^ PD901
29 | # Convert to grid
30 | grd = pygmt.xyz2grd(data=df, region=region, spacing=spacing)
   |

examples/gallery/histograms/blockm.py:48:1: PD901 Avoid using the generic variable name `df` for DataFrames
   |
46 | # Calculate number of total locations within 150x150 arc-minute bins
47 | # with blockmean's summary parameter
48 | df = pygmt.blockmean(data=data, region=region, spacing=spacing, summary="n")
   | ^^ PD901
49 | grd = pygmt.xyz2grd(data=df, region=region, spacing=spacing)
   |

examples/gallery/seismology/velo_arrow_ellipse.py:18:1: PD901 Avoid using the generic variable name `df` for DataFrames
   |
17 | fig = pygmt.Figure()
18 | df = pd.DataFrame(
   | ^^ PD901
19 |     data={
20 |         "x": [0, -8, 0, -5, 5, 0],
   |

examples/gallery/symbols/points_categorical.py:20:1: PD901 Avoid using the generic variable name `df` for DataFrames
   |
19 | # Load sample penguins data
20 | df = pd.read_csv("https://github.com/mwaskom/seaborn-data/raw/master/penguins.csv")
   | ^^ PD901
21 | 
22 | # Convert 'species' column to categorical dtype
   |

examples/tutorials/advanced/date_time_charts.py:268:1: PD901 Avoid using the generic variable name `df` for DataFrames
    |
266 |     ["20200729", 1634],
267 | ]
268 | df = pd.DataFrame(data, columns=["Date", "Score"])
    | ^^ PD901
269 | df.Date = pd.to_datetime(df["Date"], format="%Y%m%d")
    |

pygmt/clib/conversion.py:95:17: PD011 Use `.to_numpy()` instead of `.values`
   |
93 |     # East-West, North-South.
94 |     for dim in grid.dims[::-1]:
95 |         coord = grid.coords[dim].values
   |                 ^^^^^^^^^^^^^^^^^^^^^^^ PD011
96 |         coord_incs = coord[1:] - coord[0:-1]
97 |         coord_inc = coord_incs[0]
   |

pygmt/tests/test_blockmedian.py:39:13: PD011 Use `.to_numpy()` instead of `.values`
   |
37 |     a matrix.
38 |     """
39 |     table = dataframe.values
   |             ^^^^^^^^^^^^^^^^ PD011
40 |     output = blockmedian(data=table, spacing="5m", region=[245, 255, 20, 30])
41 |     assert isinstance(output, pd.DataFrame)
   |

pygmt/tests/test_datasets_earth_vertical_gravity_gradient.py:24:12: PD003 `.isna` is preferred to `.isnull`; functionality is equivalent
   |
22 |     npt.assert_allclose(data.min(), -137.125, atol=1 / 32)
23 |     npt.assert_allclose(data.max(), 104.59375, atol=1 / 32)
24 |     assert data[1, 1].isnull()
   |            ^^^^^^^^^^^^^^^^^ PD003
   |

pygmt/tests/test_datasets_samples.py:192:12: PD003 `.isna` is preferred to `.isnull`; functionality is equivalent
    |
190 |     npt.assert_allclose(grid.min(), -4929.5)
191 |     # Test for the NaN values in the remote file
192 |     assert grid[2, 21].isnull()
    |            ^^^^^^^^^^^^^^^^^^ PD003
    |

pygmt/tests/test_select.py:40:12: PD011 Use `.to_numpy()` instead of `.values`
   |
38 |     Also testing the reverse (I) alias.
39 |     """
40 |     data = dataframe.values
   |            ^^^^^^^^^^^^^^^^ PD011
41 |     output = select(data=data, region=[245.5, 254.5, 20.5, 29.5], reverse="r")
42 |     assert isinstance(output, pd.DataFrame)
   |

pygmt/tests/test_surface.py:96:12: PD011 Use `.to_numpy()` instead of `.values`
   |
94 |     Run surface by passing in a numpy array into data.
95 |     """
96 |     data = data.values  # convert pandas.DataFrame to numpy.ndarray
   |            ^^^^^^^^^^^ PD011
97 |     output = surface(
98 |         data=data,
   |

pygmt/tests/test_surface.py:135:12: PD011 Use `.to_numpy()` instead of `.values`
    |
133 |     Run surface with the -Goutputfile.nc parameter.
134 |     """
135 |     data = data.values  # convert pandas.DataFrame to numpy.ndarray
    |            ^^^^^^^^^^^ PD011
136 |     with GMTTempFile(suffix=".nc") as tmpfile:
137 |         output = surface(
    |

Found 13 errors.

Related to #2741

[weiji14]

maybe we should ignore PD901

Yeah, let's ignore PD901. I feel like df is quite a common name for pandas dataframes.

Some tests failing

[weiji14]

Need to change isna() back to isnull() for checks on xarray.DataArray. Seems like a bug on ruff not actually checking that the object is from pandas instead of xarray.

[seisman]

Seems like a bug on ruff not actually checking that the object is from pandas instead of xarray.

I have reported the bug to upstream astral-sh/ruff#8846.